Backing Up a Truck and Trailer Uisng Sets of Three-Neuron Controllers

摘要

This work is a continuation of the effort reported in several previous papers presented at ANNIE Conferences (Alexander 1994) (Alexander and Bradley 1995) by the first author. A little-modeled property of biological neurons - the ability of each neuron to fire both above and below its average firing rate - is the feature which oru model exploits. The equations developed by the author in his dissertation model neurons possessing an average firing rate. In previous papers we had demonstrated the userfulness of this feature. Thus far, these equations have successfully maintained, at constant setpoint, the height of water in a tank under conditions of randomly changing inflow, and backed a truck to a loading dock. In this paper we demonstrate the ability of another sets of equations which, also employ the average firing rate concept, to successfully back a truck with a trailer attached to a loading dock. Our algorithm for determining the steering angle (for the cab's or truck's wheels) is considerably simpler than the well-known one given by Kosko (kosko 1992).